Determination of Pepsin Digestability in Fish Meal 2000-1 PDF
Determination of Pepsin Digestability in Fish Meal 2000-1 PDF
Determination of Pepsin Digestability in Fish Meal 2000-1 PDF
April 2000
STRICTLY CONFIDENTIAL
RING TEST FOR DETERMINATION OF PEPSIN
DIGESTIBILITY IN FISH MEAL
by E.L. Miller, A.P. Bimbo, D.E. Walters, S.M. Barlow, B. Sheridan
Summary
A proposed new method to determine dilute pepsin digestibility, using a pepsin strength very
much lower than the AOAC Official Method, was tested in 16 laboratories with 12 samples of
fish meal . Results were calculated according to two different procedures in common use, i)
according to AOAC Official method 971.09 or ii) according to a method first described by
Lovern et al (1964) at the Torry Research Station, Aberdeen, Scotland. Variations in method of
shaking and source of pepsin were also investigated. Pepsin digestibility values were lower and
more variable when calculated by the Torry procedure. The Torry procedure required an
additional N determination and was more subject to other analytical errors. The method of
shaking affected the result when calculated according to the Torry but not the AOAC method.
The source of pepsin had no significant effect. A suitable dilute pepsin method has been
standardised and shown to have good repeatability and reproducibility when calculated
according to that used in the AOAC method. The type of shaker and source of pepsin are
recommended but need not be mandatory. It is recommended that this new method, with
inclusion of a blank determination to the circulated method, be adopted as an IFOMA official
method. The repeatability limit of this new method is 1.6 % units of digestibility. The
reproducibility limit is 3.3 % units of digestibility. A draft of the new recommended method is
given in Appendix 3.
Purpose
The purpose of the ring test was to establish a recommended IFOMA method for measuring
pepsin digestibility in fish meals.
Introduction
Different methods are in use to measure the pepsin digestibility of fish meals. The AOAC
Official Method 971.09 uses a strong solution of pepsin (0.2%) and gives high values of
digestibility but does not distinguish between fish meals of different quality. The Torry method
(Lovern et al 1964, Lovern 1965) uses a dilute pepsin solution (0.0002%). However, a survey of
current practice revealed a number of modifications of the Torry method were in use (Bimbo,
1998). Early studies with the Torry method indicated there might be some correlation between
pepsin digestibility and protein quality (specifically NPU) and digestibility in the rat. However,
subsequent trials did not substantiate the use of dilute pepsin digestibility as an indicator of
protein quality as measured in rat and chick growth assays (Barlow, 1976). Consequently, the
method cannot be used as a measure of the nutritional quality of protein for specific species of
animals. However, in certain circumstances fish meal traders have deemed it useful to specify
minimum digestibility levels in contracts using pepsin concentrations more dilute than that
specified by AOAC Official Method 971.09. Consequently there is a need for a standardised
dilute pepsin method of known repeatability and reproducibility for use in contracts.
1
From a review of current practices (Bimbo, 1998) a number of potential factors affecting
variability were identified and where possible (e.g. strength of pepsin, sample particle size)
appropriate standard conditions were selected as the basis of the method to be tested. Three
variables were selected for further study within the collaborative study.
(a) The AOAC method calculates pepsin digestibility as the portion of the total nitrogen in the
sample that is soluble in acid pepsin solution. The Torry method calculates pepsin digestibility
as the portion of the acid insoluble nitrogen that is soluble in acid pepsin solution. The acid
insoluble N may be substantially less than the total N. Differences in method of calculation
between laboratories would be a significant source of variation. The current study calculates the
data in these two ways.
(b) The source of pepsin is not specified in AOAC. In the present study pepsin from Merck
1:10,000 activity (Product No 7190) was specified while laboratories not using this source were
asked to standardise their source of pepsin according to ISO 6655 (1997). The effect of source
of pepsin on the results was determined.
(c) Laboratories have different shaking equipment which can be classified into either end over
end rotating as described in the AOAC method or the more readily available orbital or
reciprocating shakers. The effect of type of shaker was determined.
Method
IFOMA collected ten fish meal samples from Denmark, Norway, Chile, Peru, USA and UK.
These were the same samples as used in a ring test of determination of biogenic amines. The
samples were coded 1 to 12 and circulated by Fish Industries to 16 participating laboratories.
Samples were not reground but this was left to each participating laboratory as part of the whole
procedure. Samples 1 and 5 were a pair of hidden duplicates and samples 3 and 10 were a
second pair of hidden duplicates. Laboratories were asked to analyse each sample once only and
to report three nitrogen determinations by the kjeldahl method (ISO FDIS 5983:1997(E) for each
sample:
1. % kjeldahl nitrogen insoluble in acid pepsin solution (A)
2. % kjeldahl nitrogen insoluble in acid solution (B)
3. % kjeldahl nitrogen in the sample of fish meal (C).
In the course of conducting the ring test one laboratory (SSF) drew attention for the need to
determine and correct for the contribution of N from filter paper used to collect the insoluble
residue in determinations A and B. This was not specified in the circulated protocol and it was
possible that some laboratories would automatically include a filter paper in their reagent blank
determination of N while others might not. Participants were advised to report results without
any adjustment if they had not already completed the analysis. In a subsequent questionnaire
laboratories were asked whether or not a correction had been made and also the N supplied by a
filter paper after washing with warm distilled water. The data was analysed first as received, in
which a few had made corrections but the majority had not, and secondly after adjustment of the
reported data for the determined contribution of nitrogen from the filter paper.
2
Results
The primary measurements (A, B, C), the dilute pepsin digestibility calculated according to the
Torry method (Dig 1) and the AOAC method (Dig 2) averaged over the 12 samples are given for
each of the 16 laboratories in Table 1, together with the standard deviation determined from the
hidden duplicates (repeatability, sr) for each laboratory. A consideration of these mean values
indicates laboratories 8 and 14 returned low values for both Dig 1 and Dig 2 which can be traced
to exceptionally high values for A, the acid and pepsin insoluble residue. Adjustment of this
data for nitrogen contributed by filter paper did not account for the divergence of these two
laboratories (Table 1F).
The data were subjected to a battery of tests to determine the presence of outlier laboratories. A
Principal Component analysis was carried out using the five variable mean values displayed in
Table 1F. The Principal Component plot is shown in Figure 1. This identifies Labs 8 and 14 as
being different from the remaining group of laboratories. A Cluster Analysis examined the
grouping of laboratories in more detail using the same data. The resulting dendrogram is shown
in Figure 2. Starting with the entire set as single laboratories the laboratories are combined
either singly or in groups so as to minimise the 'Within Group' sum of squares. This quantity
starts with the value zero. The length of the horizontal line when new groupings are formed
represents the increase in the sum of squares. Figure 2 clearly shows the increased variance
when Labs 8 and 14 are added to the remainder. In addition, Labs 4, 9, 10, 15, 16 form a second
group which differ little between themselves but are different to the remaining group of Labs 1,
2, 3, 5, 6, 7, 11 and 13. Finally the data were subjected to Grubbs test (Grubbs, 1969; Horwitz,
1993). Table 2 gives the standard deviation for each of the five variables between all
laboratories and after excluding laboratory 8 and then labs 8 and 14. The Grubbs statistic, which
is the percentage reduction in standard deviation on excluding the outlying values, is also shown.
For measurement A and the derived dilute pepsin digestibilities the Grubbs statistics on
removing lab 8 and labs 8 and 14 were significantly greater than the critical values at P <0.025
(2-tail). Therefore, these two labs were regarded as outliers and the analyses are presented both
for all laboratories and after exclusion of labs 8 and 14.
The primary measurements and calculated dilute pepsin digestibilities averaged across all
laboratories are given for each meal in Table 3 and after correction for filter paper N in Table 3F.
The amount of N in a washed filter paper varied considerably from zero to 0.82 mg N per paper
used, mean 0.25 mg N SD ±0.245. Tables 4 and 4F give the corresponding values after
excluding labs 8 and 14 as outliers. These tables also give the between laboratory standard
deviations (reproducibility, sR) for each sample. The between laboratory variability expressed as
a percentage of the sample mean (Relative standard deviation, RSDR) is large for measurement
of % N insoluble in acid pepsin because the amount of residual N is very small. The amount of
N insoluble in acid alone (B) was 60.4% (SD ± 9.64) of total N (A) resulting in a substantially
lower estimate of digestibility by the Torry calculation with an accompanying greater between
laboratory variability especially when expressed as a percentage of the lower mean value
(RSDR).
Table 5A gives the hidden duplicate standard deviation pooled across laboratories (repeatability,
sr). The filter paper correction makes no difference to the repeatability of the primary
3
measurements since both hidden duplicates are treated the same but this does have an effect on
the repeatability of digestibility calculated by the Torry method but is insignificant in the
modified AOAC calculation. Omitting labs 8 and 14 have no appreciable effect on the
repeatability estimates.
Table 5B gives the between laboratory standard deviation pooled across all samples
(reproducibility, sR). Adjusting the data for filter paper N had very little effect but gave a
numerical increase rather than the expected decrease in reproducibility. Omitting labs 8 and 14
substantially reduced the reproducibility estimates for measurement A (but not B or C) and the
derived digestibility values.
Table 5C gives the variance component (s2r) due to repeatability within laboratories and the
variance component purely due to between laboratories (s2L) estimated from the sub-set of pairs
of hidden duplicate meals. If the laboratories do not contribute any variability, over and above
the errors of determination, the laboratory component would be zero. Table 5C indicates the
laboratory component is of the same order of magnitude as the basic error of determination
within one laboratory. The reproducibility variance for a single determination made by a
randomly selected laboratory is expected to be the sum of these two components (s2R = s2r + s2L).
The estimate of s2R and hence of sR in this subset of the data is generally a little less than that
estimated from the full set shown in Table 5B.
The type of shaker used had a small effect (Table 6). Less N remained after acid pepsin (A
value) when orbital shaking was used but the difference did not achieve significance (P >0.05).
In contrast less N remained after shaking in acid alone (B value) with the rotating shaker but
again the difference was not significant. In the calculation of Torry digestibility from:
Torry digestibility = 100(B-A)/B
these two trends have a combined effect in giving a greater digestibility with orbital than with
rotating shaking (P <0.05).
Table 7 classifies the data according to the source of pepsin used, after excluding the two outlier
laboratories. Eight laboratories used the specified Merck pepsin. Three laboratories used Sigma
Pepsin A 1:10,000 Porcine. Two laboratories used Difco pepsin and one used US Biochemical
20015, 1:10494. A further lab used pepsin from Saarchem Unilab which was assayed and
adjusted to the required activity. Those using Merck were compared with the average of all the
other sources. There was no significant difference in any of the measured parameters or
calculated digestibilities.
Discussion
Table 8 presents a summary of the digestibilities calculated by the two methods after the removal
of the two outlying laboratories and correcting for filter paper nitrogen. All laboratories were
capable of determining N in fish meal with acceptable repeatability (mean RSDr 0.70%, range
0.2 to 1.4% for different labs) and reproducibility (mean RSDR 1.14%, range 0.6 to 2.2% for
different samples).
Since 0.4 of the fish meal N was soluble in acid alone, digestibility calculated by the Torry
method reflects the effect of pepsin on only 0.6 of the whole N. This results in a substantial
difference to the digestibility values as calculated by the Torry and AOAC methods. In the Torry
method, the part (0.4) that is soluble in acid alone is ignored. When the Torry digestibility value
4
is used to describe a fish meal, it makes the assumption that the acid soluble material is of the
same digestibility as the acid insoluble fraction. However, material that is soluble in dilute HCl
alone in the laboratory is also likely to be readily solubilised and digested in the animal. In
contrast, this fraction is included in the AOAC definition of pepsin digestible. Furthermore the
determination of the acid insoluble N was more variable both within and between laboratories
than either the acid pepsin insoluble or the total N. Consequently, the Torry method of
calculation not only results in lower estimates of digestible nitrogen but also a greater variability
both within and between laboratories when expressed in absolute standard deviations (Table 5B)
and even more so when expressed as relative standard deviations (Table 8). When the same
sample is analysed within one laboratory the difference in absolute value between the estimates
should not exceed the repeatability limit (Table 8) in 95% of occasions. Thus while laboratories
can distinguish between two fish meals differing in crude protein by 1.38 %units, pepsin
digestibility needs to differ by 3.05 units calculated by the Torry method but only 1.57 units by
the AOAC method. Similarly, the absolute difference between two single test results obtained
using the same method on identical test material in different laboratories by different operators
using different equipment should be less than the reproducibility limit (Table 8) in 95 % of
cases. Analyses in different laboratories of the same sample could differ by as much as 5.23 units
calculated by the Torry method and 3.38 units by the AOAC method. Equally different meals
analysed in different laboratories would have to differ by at least these amounts before they
could be considered different.
Figure 3 shows the relationship between the digestibilities calculated by the two methods for
each sample averaged over the 14 laboratories. The Torry method gives a lower but extended
scale of digestibility than the AOAC method with a good relationship between the two methods
of calculation but with slope and intercept which are greatly different from unity and zero
respectively. A good relationship is to be expected in this exercise because the same acid pepsin
insoluble N determination is used in both calculations. This relationship does not apply to
values determined by the original AOAC method (using 0.2% pepsin) compared with the usual
Torry method. The greater scale of the Torry method (a range of 14 % units, 83 to 97 in the
present samples compared with a range of 9 % units, 89 to 98) is offset by the 94% greater
within laboratory variability and 60% greater between laboratory variability. Consequently, the
Torry method of calculation is not more discriminatory between samples. In addition the Torry
method requires an extra N determination of the acid insoluble N and is more affected by
problems of filter paper N contamination of the filtered residues. Consequently, the AOAC
method of calculation is preferred and is adopted for the future method.
The type of shaking equipment used affected the digestibility calculated according to the Torry
method but not that by the modified AOAC method. Although the differences in the A and B
values were not significant they suggest that the enzyme digest proceeded best under the gentle
conditions of orbital shaking and that failure to digest sample through material being deposited
on the vessel walls out of the solution was not a problem. In contrast solubility in dilute HCl
alone seems to have been enhanced by the more vigorous shaking conditions in the end over end
rotating shaker. The difference due to shaker type is a further reason for preferring the AOAC
method of calculating results. No recommendation need be made to use a particular type of
shaker but the preference would be for the more readily available modern orbital type of shaker.
The source of pepsin did not apparently affect the results. However, laboratory 12 reported the
lowest amount of acid pepsin insoluble N and also used pepsin from Saarchem Unilab which
required adjustment of the amount used to achieve the desired activity. It is conceivable that too
5
much activity was used. The two outlier laboratories reporting very high amounts of acid pepsin
insoluble N both used Sigma pepsin but three other laboratories using the same pepsin source
reported values within the range of the Merck enzyme. Consequently, it does seem prudent to
specify the use of one source of enzyme.
Appendix 3 details the new method to be adopted by IFOMA as its official method. The type of
shaker and source of pepsin are recommended but need not be mandatory. While the adoption of
a common method may help reduce apparent discrepancies between laboratories apparently
reporting the same analysis attention is drawn to the quite large repeatability and reproducibility
limits (Table 8 and Appendix 3) relative to the range of values (88 – 98) likely in normal
commerce. Thus, within one laboratory differences between fish meals of less than 1.6 %
digestibility units cannot be distinguished. When the same fish meal is analysed in two different
laboratories differences of up to 3.3 % digestibility units are to be expected.
References
AOAC (1999). Official method 971.09. Official Methods of Analysis of AOAC International,
16th edition 5th revision. AOAC International: Gaithersburg, MD 20877-2417, USA.
Barlow, S.M. (1976). Pepsin digestibility as an index of fish meal quality – Part 3. Fishing News
International, 15:25, 26, 29.
Bimbo, A.P. (1998). An evaluation of the pepsin digestibility method as practised and
interpreted by IFOMA members, associate members and associated laboratories. IFOMA
Research Report 1998-1. International Fishmeal & Oil Manufacturers Association.
Grubbs, F.E. (1969). Procedures for detecting outlying observations in samples. Technometrics
11: 1-21.
Horwitz, W. (1993). Protocol for the design, conduct and interpretation of collaborative studies.
Report of International Union of Pure and Applied Chemistry.
ISO 6655 (1997). Animal feedingstuffs – Determination of the soluble nitrogen content after
treatment with pepsin in hydrochloric acid solution. Annex A. Determination of pepsin activity.
Lovern, J.A. (1965). Some analytical problems in the analysis of fish and fish products. Journal
of the A.O.A.C. 48: 60-68.
Lovern, J.A., Olley, J. and Pirie, R. (1964). Fishing News International, 3: 310.
6
Table 1 Table giving the Laboratory Mean Values, and the 'hidden duplicate' Standard
Deviation for each Laboratory, and variable. (No correction for filter paper N.)
1 0.587 6.98 11.14 91.38 94.63 0.059 0.022 0.099 1.025 0.484
2 0.569 6.92 11.17 91.63 94.81 0.036 0.074 0.141 0.585 0.276
3 0.584 7.01 11.18 91.51 94.68 0.125 0.032 0.018 2.047 1.068
4 0.415 6.65 11.26 93.65 96.24 0.010 0.250 0.021 0.150 0.083
5 0.531 6.01 11.10 91.09 95.11 0.029 0.406 0.092 0.822 0.276
6 0.592 6.54 11.31 90.84 94.65 0.100 0.175 0.025 1.567 0.861
7 0.543 6.42 11.30 91.28 95.10 0.043 0.086 0.040 0.867 0.375
8 1.007 6.85 11.19 84.92 90.89 0.043 0.040 0.055 0.843 0.339
9 0.378 6.57 11.13 94.14 96.54 0.000 0.032 0.057 0.038 0.011
10 0.427 6.98 11.06 93.61 96.04 0.021 0.074 0.050 0.442 0.172
11 0.538 6.27 11.12 91.06 95.09 0.150 0.525 0.039 2.416 1.316
12 0.350 6.91 11.21 94.73 96.84 0.011 0.101 0.160 0.298 0.132
13 0.573 7.06 11.19 91.68 94.77 0.049 0.084 0.073 0.729 0.434
14 0.814 7.01 11.08 88.13 92.53 0.057 0.110 0.010 1.191 0.506
15 0.508 6.84 11.12 92.38 95.33 0.010 0.059 0.046 0.161 0.100
16 0.477 6.88 11.22 92.89 95.65 0.021 0.025 0.066 0.370 0.158
Table 1F Table giving the Laboratory Mean Values, and the ‘hidden duplicate’ Standard
Deviation for each Laboratory, and variable. (Corrected for filter paper N.)
7
Table 2. Standard deviation between laboratories for N soluble in acid pepsin (A), N soluble in
acid alone (B), total N (C) and the dilute pepsin digestibilities calculated according to the Torry
(Dig. 1) and AOAC (Dig. 2) procedures. The Grubbs statistic is also given (in brackets) where
significant. The * denotes statistical significance at the P <0.025 level.
8
Table 3. Sample means, 'Between Laboratories' Standard Deviations, and Relative Standard
Deviations. (No correction for filter paper N.)
Sample 1 2 3 4 5 6 7 8 9 10 11 12
Mean 0.340 0.378 0.527 0.520 0.343 0.301 0.557 0.451 0.861 0.515 1.135 0.731
S.D. 0.147 0.153 0.188 0.256 0.140 0.123 0.196 0.163 0.269 0.180 0.188 0.236
RSDR 43.31 40.47 35.78 49.27 40.78 40.76 35.16 36.22 31.22 34.94 16.60 32.21
Mean 6.036 7.661 5.388 7.619 6.006 8.110 5.347 7.761 7.351 5.426 6.500 7.718
S.D. 0.560 0.500 0.246 0.317 0.535 0.435 0.191 0.594 1.059 0.162 0.249 0.244
RSDR 9.28 6.53 4.56 4.15 8.92 5.37 3.57 7.66 14.41 2.98 3.83 3.16
Mean 11.74 11.39 11.34 11.18 11.77 11.52 10.54 11.27 11.00 11.32 9.98 11.01
S.D. 0.110 0.137 0.087 0.117 0.110 0.087 0.222 0.113 0.073 0.128 0.155 0.077
RSDR 0.93 1.20 0.77 1.04 0.93 0.76 2.10 1.01 0.67 1.13 1.56 0.70
Mean 94.31 95.11 90.27 93.20 94.25 96.28 89.59 94.13 88.18 90.54 82.53 90.53
S.D. 2.527 1.839 3.279 3.244 2.368 1.515 3.584 2.267 3.349 3.141 2.843 2.985
RSDR 2.68 1.93 3.63 3.48 2.51 1.57 4.00 2.41 3.80 3.47 3.45 3.30
Mean 97.11 96.68 95.36 95.36 97.09 97.38 94.70 96.00 92.18 95.45 88.61 93.35
S.D. 1.241 1.338 1.640 2.273 1.181 1.076 1.909 1.457 2.439 1.583 1.969 2.159
RSDR 1.28 1.38 1.72 2.38 1.22 1.11 2.02 1.52 2.65 1.66 2.22 2.31
9
Table 3F. Sample means, 'Between Laboratories' Standard Deviations, and Relative Standard
Deviations. (Corrected for filter paper N)
Sample 1 2 3 4 5 6 7 8 9 10 11 12
Mean 0.323 0.362 0.509 0.503 0.326 0.284 0.540 0.434 0.843 0.498 1.118 0.714
S.D. 0.149 0.157 0.189 0.262 0.147 0.124 0.203 0.169 0.276 0.187 0.198 0.241
RSDR 46.32 43.25 37.02 52.20 45.12 43.55 37.55 38.99 32.68 37.49 17.74 33.83
Mean 6.018 7.646 5.371 7.601 5.988 8.093 5.330 7.744 7.333 5.408 6.483 7.701
S.D. 0.560 0.497 0.241 0.314 0.542 0.433 0.187 0.592 1.069 0.164 0.252 0.250
RSDR 9.31 6.50 4.48 4.14 9.04 5.35 3.51 7.64 14.58 3.04 3.89 3.24
Mean 11.74 11.39 11.34 11.18 11.77 11.52 10.54 11.27 11.00 11.32 9.98 11.01
S.D. 0.110 0.137 0.087 0.117 0.110 0.087 0.222 0.113 0.073 0.128 0.155 0.077
RSDR 0.93 1.20 0.77 1.04 0.93 0.76 2.10 1.01 0.67 1.13 1.56 0.70
Mean 94.59 95.31 90.57 93.42 94.53 96.49 89.88 94.34 88.43 90.84 82.75 90.74
S.D. 2.569 1.892 3.293 3.326 2.471 1.525 3.715 2.339 3.317 3.253 2.970 3.053
RSDR 2.72 1.99 3.64 3.56 2.61 1.58 4.13 2.48 3.75 3.58 3.59 3.36
Mean 97.26 96.82 95.51 95.51 97.24 97.53 94.86 96.15 92.34 95.61 88.79 93.51
S.D. 1.258 1.368 1.640 2.328 1.242 1.083 1.972 1.507 2.502 1.642 2.064 2.212
RSDR 1.29 1.41 1.72 2.44 1.28 1.11 2.08 1.57 2.71 1.72 2.33 2.37
10
Table 4. Sample means and 'Between Laboratories' Standard Deviations, and Relative Standard
Deviations. (Excluding Labs. 8 & 14 as outliers). (No Correction for filter paper N.)
Sample 1 2 3 4 5 6 7 8 9 10 11 12
Mean 0.298 0.328 0.478 0.459 0.304 0.263 0.493 0.401 0.790 0.460 1.086 0.686
S. D. 0.086 0.075 0.095 0.208 0.093 0.066 0.091 0.092 0.158 0.076 0.143 0.215
RSDR 28.71 22.98 19.94 45.34 30.71 24.99 18.36 22.91 20.04 16.49 13.16 31.33
Mean 6.031 7.608 5.358 7.605 6.009 8.089 5.341 7.708 7.280 5.406 6.467 7.706
S. D. 0.595 0.514 0.246 0.335 0.573 0.450 0.186 0.618 1.119 0.163 0.235 0.252
RSDR 9.87 6.75 4.60 4.40 9.53 5.57 3.48 8.02 15.37 3.02 3.64 3.27
Mean 11.74 11.40 11.34 11.17 11.78 11.53 10.56 11.28 11.01 11.32 10.00 11.02
S. D. 0.107 0.140 0.084 0.122 0.105 0.069 0.232 0.120 0.077 0.136 0.156 0.075
RSDR 0.91 1.23 0.74 1.09 0.89 0.60 2.19 1.07 0.70 1.20 1.56 0.68
Mean 95.01 95.70 91.12 93.99 94.90 96.74 90.77 94.70 88.96 91.50 83.16 91.09
S. D. 1.512 0.950 1.551 2.602 1.575 0.855 1.631 1.724 2.292 1.346 2.418 2.748
RSDR 1.59 0.99 1.70 2.77 1.66 0.88 1.80 1.82 2.58 1.47 2.91 3.02
Mean 97.46 97.12 95.79 95.90 97.42 97.72 95.32 96.44 92.83 95.94 89.12 93.77
S. D. 0.723 0.663 0.840 1.848 0.781 0.574 0.897 0.813 1.424 0.658 1.484 1.960
RSDR 0.74 0.68 0.88 1.93 0.80 0.59 0.94 0.84 1.53 0.69 1.67 2.09
11
Table 4F. Sample means and 'Between Laboratories' Standard Deviations, and Relative
Standard Deviations. (Excluding Labs. 8 & 14 as outliers.) (Corrected for filter paper N.)
Sample 1 2 3 4 5 6 7 8 9 10 11 12
Mean 0.278 0.310 0.458 0.439 0.285 0.243 0.473 0.382 0.770 0.440 1.067 0.667
S.D. 0.080 0.072 0.084 0.211 0.097 0.055 0.094 0.093 0.163 0.079 0.151 0.219
RSDR 28.59 23.14 18.33 48.06 34.22 22.57 19.83 24.37 21.10 17.99 14.20 32.79
Mean 6.012 7.591 5.338 7.585 5.989 8.069 5.321 7.688 7.260 5.386 6.447 7.687
S.D. 0.595 0.508 0.239 0.331 0.579 0.447 0.180 0.614 1.128 0.163 0.237 0.258
RSDR 9.90 6.69 4.47 4.37 9.67 5.54 3.39 7.98 15.54 3.03 3.67 3.35
Mean 11.74 11.40 11.34 11.17 11.78 11.53 10.56 11.28 11.01 11.32 10.00 11.02
S.D. 0.107 0.140 0.084 0.122 0.105 0.069 0.232 0.120 0.077 0.136 0.156 0.075
RSDR 0.91 1.23 0.74 1.09 0.89 0.60 2.19 1.07 0.70 1.20 1.56 0.68
Mean 95.32 95.92 91.44 94.23 95.22 96.97 91.11 94.94 89.24 91.83 83.42 91.33
S.D 1.442 0.930 1.386 2.647 1.618 0.740 1.726 1.754 2.130 1.405 2.523 2.786
RSDR 1.51 0.97 1.52 2.81 1.70 0.76 1.89 1.85 2.39 1.53 3.02 3.05
Mean 97.63 97.28 95.96 96.08 97.59 97.89 95.51 96.62 93.00 96.11 89.32 93.95
S.D. 0.671 0.627 0.739 1.876 0.816 0.479 0.926 0.821 1.464 0.690 1.566 1.992
RSDR 0.69 0.64 0.77 1.95 0.84 0.49 0.97 0.85 1.57 0.72 1.75 2.12
12
Table 5A. Pooled 'Within Laboratory' standard deviations from the hidden duplicates, both for
full data set, and after removing outliers (Labs 8 & 14), WITH and WITHOUT correction for N
contributed by filter paper.
.
A B C Dig.1 Dig.2
All Laboratories
Without N Adj. 0.063 0.192 0.075 1.073 0.547
With N Adj. 0.063 0.192 0.075 1.083 0.547
Table 5B. Pooled 'Between Laboratory' standard deviations, both for full data set, and after
removing outliers (Labs 8 & 14), WITH and WITHOUT correction for N contributed by filter
paper.
A B C Dig.1 Dig.2
All Laboratories
Without N Adj. 0.192 0.490 0.125 2.818 1.746
With N Adj. 0.197 0.491 0.125 2.884 1.795
Table 5C. Pooled estimates of the components of variance derived from the subset of data of the
hidden duplicate samples of fish meals, after excluding outliers (Labs 8 and 14) and after
correction for N contributed by filter paper.
13
Table 6. Table displaying difference in variables due to Type of Shaker. Reduced Data set,
excluding Labs 8 & 14, and after correction for N contributed by filter paper.
Table 7. Table displaying difference in variables due to Pepsin Source. Reduced Data set,
excluding Labs 8 & 14, and after correction for filter paper N.
Table 8. Mean overall samples of fish meal N and dilute pepsin digestibility calculated by two
methods (Torry and modified AOAC) together with the relative repeatability and reproducibility
standard deviations. Reduced Data set, excluding Labs 8 & 14, and after correction for filter
paper N.
14
Figure 1 Plot of first two Principal Components. Labs 8 and 14 are seen to be separate from the
remaining population of laboratories.
1.5
Lab 14
Second Principal Component
0.5
-0.5 Lab 8
-1
-1.5
-2
-2.5
-3 -2 -1 0 1 2
First Prinicpal Component
15
Figure 2.Dendrogram illustrating the grouping of laboratories. The length of the horizontal line
joining groups is proportional to the increase in within group sums of squares when the lab or
group of labs are combined.
16
Figure 3. Relationship between dilute pepsin digestibility of N of 12 samples of fish meal
calculated according to either the AOAC or Torry method from the same primary data.
100
98
y = 0.6324x + 37.031
AOAC method (%)
96 R2 = 0.922
94
92
90
88
82 84 86 88 90 92 94 96 98
17
Appendix 1. The circulated protocol.
1.0 Applicable to fishmeal. This method cannot be used as a measure of the nutritional quality
of protein for specific species of animals. However, in certain circumstances fish meal traders
have deemed it useful to specify minimum digestibility levels in contracts using pepsin
concentrations more dilute than that specified by AOAC Official Method 971.09. The purpose
of this method is to harmonise the dilute pepsin digestibility procedure.
3.0 Summary: Protein is digested according to the conditions of the test. Determine the
kjeldahl nitrogen insoluble in pepsin/acid solution, the kjeldahl nitrogen insoluble in the acid
solution and the kjeldahl nitrogen in the original fishmeal sample.
The portion of the acid insoluble nitrogen that is soluble in pepsin/acid solution.
The portion of the total nitrogen in the sample that is acid/pepsin soluble.
5.0 Reagents
1. Merck pepsin 1:10,000 activity (Product No. 7190) (See note 1). Store cool and dry.
2. Hydrochloric Acid, 37%.
3. Distilled water
4. 0.075 N Hydrochloric Acid. Dilute 6.3 ml of 37% hydrochloric acid to 1000 ml, or 12.6 ml
to 2000 ml with distilled water.
5. 0.020% Pepsin/Acid Solution. Prepare fresh.
Dissolve 0.2000 grams of 1:10,000 activity pepsin in a litre of 0.075 N Hydrochloric Acid
6. 0.0002% Pepsin/Acid Solution. Prepare fresh.
Transfer 20 ml of the 0.020% pepsin solution to a 2 litre volumetric flask and dilute to
18
the mark with 0.075 N Hydrochloric Acid.
7. Also, apparatus and chemicals needed for the determination of kjeldahl protein. (ISO/FDIS
5983:1997(E).
6.0 Procedure
1. Grind the sample so that the particles pass through an ASTM No. 30-32 Screen
(0.500-0.595mm). If it is difficult to grind to this particle size, then the fat content may be too
high and the sample should be extracted with petroleum ether for 2 hours. Extraction should be
done on the 1 gram weighed sample and the residue used for the rest of the test.
2. Weigh accurately 1.0 gram and record the weight to 0.1 mg, of the ground fishmeal into the
200 ml reaction bottle or flask. Mark this container as pepsin.
3. Weigh accurately 1.0 gram (and record weight to 0.1 mg) of the ground fishmeal into a
second 200 ml reaction bottle or flask. Mark this container as acid.
4. Transfer 150 ml of the 0.0002% pepsin/acid solution (5.6) into the flask marked "pepsin".
5. Transfer 150 ml of the 0.075 N. Hydrochloric Acid Solution (5.4) into the flask marked
"acid".
6. Incubate the flasks for 16 hours at 45°C±2.0. If an orbital shaker unit is used be sure there
are no particles on the wall of the flask above the pepsin/acid solution. The orbital shaker
should shake at 140-200 orbits/minute.
7. Filter the solutions through the filter paper using a gentle vacuum suction. Wash out any
remaining particles from the flask or bottle with warm distilled water and transfer to the filter.
Finally wash the filter with several portions of warm distilled water.
8. Determine the kjeldahl nitrogen content of the filtered residues according to
ISO/FDIS 5983:1997(E).
9. Determine the kjeldahl nitrogen content of the original fishmeal sample according to
ISO FDIS 5983:1997(E).
7.0 CALCULATIONS
19
Data is to be reported to 2 decimal places.
The recommended method for determination of pepsin activity is given in Annex A (3 pages
attached) taken from ISO 6655 (1997).
20
Appendix 2. Participants in dilute pepsin digestibility ring test.
Analyst Laboratory
21
Lima 27, Peru
Arne Brodin SSF
Kjerreidviken 16,
Fyllingsdalen N-5033,
Norway
Gunnhildur Gísladóttir The Icelandic Fisheries Laboratory,
PO Box 1405,
Skulagata 4,
121 Reykjavik, Iceland
Boyce H. Butler Thionville Laboratories Inc.
PO Box 23687, 5440 Pepsi Street,
New Orleans,
LA 70183, USA
22
APPENDIX 3
1.0 Applicable to fishmeal. This method cannot be used as a measure of the nutritional
quality of protein for specific species of animals. However, in certain circumstances fish meal
traders have deemed it useful to specify minimum solubility levels in contracts using pepsin
concentrations more dilute than that specified by AOAC Official Method 971.09. The purpose
of this method is to harmonise the dilute pepsin solubility procedure.
3.0 Summary: Protein is digested according to the conditions of the test. Determine the
Kjeldahl nitrogen insoluble in pepsin/acid solution and the Kjeldahl nitrogen in the original
fishmeal sample.
The portion of the total nitrogen in the sample that is pepsin/acid soluble.
5.0. Reagents
5.1. Merck pepsin 2,000 FIP-U/g activity (Product No. 107190) (See note 1). Store cool and
dry.
5.2. Hydrochloric acid, 37%.
5.3. Distilled water
5.4. 0.075 M Hydrochloric acid. Dilute 6.3 ml of 37% hydrochloric acid to 1000 ml, or 12.6 ml
to 2000 ml with distilled water.
5.5. 0.020% Pepsin/acid solution. Prepare fresh.
Dissolve 0.2000 grams of 1:10,000 activity pepsin in a litre of 0.075 N. Hydrochloric acid
5.6. 0.0002% Pepsin/acid solution. Prepare fresh.
Transfer 20 ml of the 0.020% pepsin solution to a 2 litre volumetric flask and dilute to
the mark with 0.075 M Hydrochloric acid.
5.7. Also, apparatus and chemicals needed for the determination of Kjeldahl nitrogen
(ISO/FDIS 5983:1997(E).
23
6.0. Procedure
6.1. Grind the sample so that the particles pass through an ASTM No. 30-32 Screen
(0.500-0.595mm). If it is difficult to grind to this particle size, then the fat content may be too
high and the sample should be extracted with petroleum ether for 2 hours. Extraction should be
done on the 1 gram weighed sample and the residue used for the rest of the test.
6.2. Weigh accurately 1.0 gram and record the weight to 0.1 mg, of the ground fishmeal into the
200 ml reaction bottle or flask.
6.3. Add 150 ml of the 0.0002% pepsin/acid solution (5.6).
6.4. Incubate the flask for 16 hours at 45°C±2.0. If an orbital shaker unit is used be sure there
are no particles on the wall of the flask above the pepsin/acid solution. The orbital shaker
should shake at 140-200 orbits/minute.
6.5. Filter the solution through the filter paper on the Buchner funnel using a gentle vacuum
suction. Wash out any remaining particles from the flask or bottle with warm distilled water and
transfer to the filter. Finally wash the filter and residue with several portions of warm distilled
water until they are free from acid.
6.6. Determine the Kjeldahl nitrogen in the filtered residues plus filter paper (Nr) according to
ISO/FDIS 5983:1997(E).
6.7. Determine the Kjeldahl nitrogen content of the original fishmeal sample according to
ISO/FDIS 5983:1997(E).
8.0. Calculations
24
9.0. Precision
The results of an inter-laboratory test on the precision of the method are summarised below:
Sixteen laboratories analysed 12 samples of fish meal. Two pairs of hidden duplicate meals were
included from which repeatability was determined. After omitting two laboratories as outliers
the sample mean solubilities ranged from 89.3 to 97.9, mean 95.6, standard deviation between
sample means 2.46. The pooled variance component for within laboratory differences (s2r) was
0.315, the repeatability standard deviation (sr) was determined as 0.56 % units and the relative
repeatability standard deviation (RSDr) as 0.59 %. The pooled variance component for between
laboratory differences (s2R) was 0.234 (from the hidden pairs only) providing an estimate of the
pure between laboratory standard deviation (sL) of 0.484 % units. The reproducibility standard
deviation (sR, estimated from comparison between laboratories of all 12 samples) was 1.17 %
units and the relative reproducibility standard deviation (RSDR) as 1.23. Source of pepsin and
type of shaker had no significant effect on the determination.
The absolute difference between two independent single test results obtained by the same
method on identical test material in the same laboratory by the same operator using the same
equipment within a short interval of time should be less than the repeatability limit (r = 2.8 sr) of
1.6 % units in 95% of cases.
The absolute difference between two single test results obtained using the same method on
identical test material in different laboratories by different operators using different equipment
should be less than the reproducibility limit (R = 2.8 sR) of 3.3 % units in 95 % of cases.
Note 1:
If other brands of pepsin (see 5.1) are used it is essential to confirm their activity at 2,000 FIP
u/g. Any variations should be adjusted when preparing the dilute pepsin solution (see 5.5 and
5.6).
The recommended method for determination of pepsin activity is given in Annex A of ISO 6655
(1997).
25